This Has My Attention: New Questions

So I know you said that you don't trust data you can't see, but I don't trust anecdotal evidence - so I again looked at my data of close to 12,000 teams and find that the average pythagorean error is 0 and the standard deviation is 4.3. The percentage of teams with errors at +4.3 or more is 16% and the percentage of teams at -4.3 or less is 16%.

"The work of science is to substitute facts for appearances and demonstrations for impressions" - John Ruskin

Wow

So I know you said that you don't trust data you can't see, but I don't trust anecdotal evidence - so I again looked at my data of close to 12,000 teams and find that the average pythagorean error is 0 and the standard deviation is 4.3. The percentage of teams with errors at +4.3 or more is 16% and the percentage of teams at -4.3 or less is 16%.

"The work of science is to substitute facts for appearances and demonstrations for impressions" - John Ruskin

So cool. Is there any way to analyze for managers instead of teams?

edit: never mind, you already did that on page 6 haha

It looks like I don’t have anything to worry about then. It just needs some more time - but it’s definitely been no fun being deep in the negative tail.

I wanna get me some of this:

Degrom strikes out 23 of 24 outs in 8 innings:

https://365.strat-o-matic.com/game/463015/950

Pulled for relief who lost the game in the 9th!

Thinking about analyzing 12,000 teams…

The average error is REQUIRED to be zero, is it not? It can’t be otherwise, so other than then the +/- 4.3, what is that telling us? Also, 12,000 teams: recent, old, before/after Bullpen 3, ATG9, all player sets, etc etc etc?

Lots of devils in the details.

Science does not require trust. It requires understanding.

You know what...I just have fun playing. At my age getting into all of this conjecture/math analysis, not only is beyond me, but takes away from the enjoyment of the game. What's wrong with trying to beat the system as it is. Anyway I'm still learning a lot from some of you experts and I thank you for your help.

You know what...I just have fun playing. At my age getting into all of this conjecture/math analysis, not only is beyond me, but takes away from the enjoyment of the game. What's wrong with trying to beat the system as it is. Anyway I'm still learning a lot from some of you experts and I thank you for your help.

Could not agree more. I am now into my third year and this may be the most frustrating endeavor (for fun & entertainment) I have tried. I am not a mathematician or big into statistics. I just have always loved baseball and the art of competition (and as an old timer I hate what has become of the real game). I see other players like Maxpower (and many others) and am in awe as to how they mange to win way more than they lose but that is all part of the fun, is trying to figure out how to be as successful but at the same time to just enjoy and have some fun. I have no idea if there is some nefarious game being played behind the curtain that with some STRAT wizard attempting to alter the results of games. Honestly, I don't know if I really care. If it meant that much to me I guess I would not be playing. After all it is my choice. At some point I may say I have had enough, the game has run its course and it is time to leave and find something else, or maybe not and I will continue for many more years trying to figure out how the hell to win at this crazy game. And, yes most of all it is just a game.

Thinking about analyzing 12,000 teams…

The average error is REQUIRED to be zero, is it not?

No it is not required to be zero, but yes, an unbiased estimator would have an average error of 0

It can’t be otherwise, so other than then the +/- 4.3, what is that telling us?

I'm describing the parameters of the distribution of errors. You've mentioned a normal distribution and I'm saying that, from this data, the errors appear to be normally distributed with mean=0 and sd=4.3. More importantly I'm saying that there are as many on the + 1 sd side as on the - 1 sd side. You appeared to be questioning that. By the way, I found a study of the error of the pythagorean and other predictors for mlb and the RMSE (virtually same calculation as sd) for pythagorean-2 was 4.1. Pythagenport's RMSE was 4.0

Also, 12,000 teams: recent, old, before/after Bullpen 3, ATG9, all player sets, etc etc etc?

league numbers from 461000 through 462868, filtering out football, free trial, and other invalid numbers. So that's roughly from July 2022 through recently completed leagues. Any player set used in that time period
[/i]

One trillion runs scored minus one trillion runs allowed will ALWAYS be zero. (Substitute any number for a trillion).

So it is required. There’s no “wow” in that.

Its not required because the pythagorean is a function of the square of runs scored and runs allowed. Check out any random league and you'll see unlike actual wins and losses and run differential, the pythagorean wins and losses do not sum to zero - the sum varies from league to league. That being said, I can't see any reason a priori why the mean of the deviations barrmorris looked at should be zero. And anyway, it doesn't mean that you are incorrect in your assumption J-Pav. If high run differential teams are typically below the pythag predictions, and veteran managers like you have more than your share of good teams, you could be below the predictions more often than not.

the pythagorean wins and losses do not sum to zero - the sum varies from league to league.

You’ll need to show me that, because I’m not following you.